Results of interpretability test on sound effects generated by our model. Average interpretability for generated sound effects is 62%